96 results about "Dataflow computation" patented technology

The invention relates to the technical field of electric digital data processing, in particular to a data processing method for quality monitoring of a power distribution equipment production line, which comprises the following steps of: calculating a time decay dynamic fluctuation index based on acquired continuous power distribution equipment sensor data flow, and recording data points and timestamps of which variable quantities exceed a threshold value; and obtaining a time decay sampling point sequence. According to the dynamic sensing and time decay modeling mechanism of the continuous power distribution equipment sensor data flow, the time decay dynamic fluctuation index is constructed to capture numerical value abrupt change in a short period, key change points where quality abnormity possibly exists can be positioned in time, timestamps can be recorded, and the response sensitivity to data behaviors before and after event triggering is improved. Dense original data segments are actively intercepted after anomaly recognition, multi-stage discrete wavelet decomposition of time domain signals is carried out, smooth trend and multi-scale mutation features are extracted in a refined mode, and the depicting capacity of key process detail changes of the circuit breaker is enhanced.

The invention provides a data processing priority scheduling method for high-speed satellite communication and related equipment, which can effectively improve the transmission performance of a high-speed satellite communication system so as to adapt to differentiated requirements of various communication scenes. The method comprises the following steps: acquiring cross-layer state information; calling a strategy matrix, wherein the strategy matrix is used for defining weight coefficients of different data types under different network states and channel qualities; calculating a target priority score for the to-be-transmitted data stream according to the cross-layer state information and the strategy matrix; sequencing the target priority scores from high to low to obtain a sequencing result; scheduling transmission queues of the data streams for transmission according to a sorting result, and allocating cache transmission queues for the data streams with high priority scores; when it is monitored that the priority score of the new data stream exceeds a preset emergency threshold value, the new data stream is inserted into a cache transmission queue of the transmission queue, and back pressure control is carried out on data streams of other priorities.

The invention relates to the technical field of data processing, in particular to an auxiliary driving device and system based on video image processing, and the device comprises a frame buffer queue with a multi-channel video collection module aligned with timestamps; a hierarchical calculation task decoupling module identifies task types in the frame buffer queue and marks priority labels, including a first priority label of a radar data analysis and distance superposition task, a second priority label of a vehicle profile generation task and a third priority label of a video fusion task, and outputs a data stream with priority labels; the calculation module generates a task processing result; the genetic algorithm optimization scheduling module generates a resource allocation strategy and controls the calculation module to execute task processing; and the output synchronous control module synthesizes a high-priority task result and performs frame rate synchronous output on a low-priority task. According to the method, the problem of real-time degradation caused by multi-path high-definition video stream processing is solved through task grading and dynamic resource scheduling, and the image output delay of the auxiliary driving scene is reduced.

The invention relates to the technical field of big data preprocessing, in particular to a smart agriculture-oriented big data preprocessing system, which comprises a smart agriculture data fingerprint construction module for calculating a sliding window standard deviation and a frequency exceeding a preset crop canopy temperature threshold value for a data stream of each smart agriculture sensor, and jointly calculating a short-term correlation coefficient with the reading of the adjacent sensor. According to the method, the standard deviation and the over-limit frequency of the sliding window of the data stream of the sensor are extracted, and the short-term correlation coefficient between adjacent sensors is introduced to construct the multi-dimensional quality feature vector, so that the high-resolution recognition of the abnormal structure of the data is realized, and the accuracy of the quality evaluation of the original data is enhanced; and comparing the standard deviation and the short-term correlation coefficient with a preset threshold value, performing intersection screening, accurately defining a suspected noise data segment, dynamically calculating a predicted value of each time point in a local sequence context, realizing point-by-point repair of an abnormal segment, and retaining natural continuity of a time structure.

The invention relates to the technical field of databases, and particularly provides a streaming computing method and device for a distributed database, and the method comprises the following steps: S1, receiving time sequence data change captured by a CDC component in real time, and storing the time sequence data change in a local cache; s2, performing stream aggregation calculation of stream calculation; s3, receiving a calculation result of stream calculation, converting a result set into a target table compatible format, and writing data into a target table by using an internal actuator; s4, processing existing historical data during stream calculation creation, breakpoint data generated during a stream calculation pause period and expired data exceeding a time sequence data out-of-order window; and S5, managing the life cycle of the flow calculation in the distributed database cluster. Compared with the prior art, the method has the advantages that the streaming calculation of the time series data of the database can be efficiently completed, the recalculation of the window split data is efficiently completed by using the streaming calculation recalculation module, and the performance attenuation is within 15% of the original storage performance.

The invention provides an energy-carbon management system, a construction method of the energy-carbon management system, a calculation method of energy-carbon management data and a computer readable storage medium. The energy-carbon management system comprises a data entry port, at least one virtual node, a data processing module and a stream calculation engine. The data entry port can receive virtual data, and the virtual data cannot be directly obtained through measurement equipment. One virtual node corresponds to one piece of virtual data, and the corresponding virtual data can be obtained through the data entry port. And the data processing module communicates with the virtual nodes and can execute a preset calculation process to obtain secondary data. The stream calculation engine can obtain the secondary data from the data processing module, and can execute a preset stream calculation based on calculation data, and output energy carbon management data, the calculation data including the secondary data. According to the embodiment of the invention, the energy carbon management data can be quickly and accurately output in real time, and the manual labor is reduced.

The invention relates to the technical field of machine vision, in particular to an aluminum hose quality inspection method and device based on machine vision and computer equipment. The method comprises the following steps: configuring a defect-illumination response matrix, wherein the response matrix comprises detection confidence coefficient values corresponding to multiple preset defect types under multiple illumination conditions; acquiring a historical detection data stream on a current production line, and calculating a prediction probability vector of the multiple preset defect types at the current moment based on the historical detection data stream; inputting the prediction probability vector into an optimization model with a preset constraint condition, and combining the defect-illumination response matrix to solve a target illumination combination which maximizes a weighted detection confidence expectation value; and triggering a light source assembly to work according to the target illumination combination, collecting a surface image of the to-be-detected aluminum hose, and outputting a quality detection result based on the surface image. According to the method, the defects of the aluminum hose can be identified more accurately.

The invention relates to the technical field of intelligent automobile electronic control, and discloses a vehicle-mounted network controller refreshing system and method.The vehicle-mounted network controller refreshing system comprises a data flow classification and priority calibration module, a function key score calculation module and a function key score calculation module, discrete priorities P0 to P3 are generated through weighted mapping, and corresponding delay upper limits are marked; and the flow prediction and dynamic bandwidth allocation module is used for receiving the output priority label and the delay upper limit and smoothly predicting the bandwidth demand of the next period according to an index. According to the method, the technical scheme of weighted mapping of the timeliness score and the function criticality score is adopted, so that the technical effects of dynamically dividing the P0-P3 priorities based on the flow characteristics and binding the delay upper limit are achieved, and compared with the technical scheme of manually presetting the fixed priority or simply polling and scheduling in the prior art, the method has the advantages that the practicability is high; the defects that emergency data streams and common data streams cannot be distinguished in a self-adaptive mode, and delay constraints and bandwidth requirements are disjointed are overcome.

The invention relates to the technical field of medical care, and provides an electronic medical record-based nursing system, which comprises a heterogeneous data integration module for acquiring multi-source heterogeneous electronic medical record data from a hospital HIS (hospital information system), wearable monitoring equipment and a nursing operation terminal and outputting cleaned and standardized data; the feature association module is used for receiving the cleaned and standardized data to generate a spatial-temporal feature association matrix and storing a feature topological graph; the dynamic decision-making module is used for processing the feature incidence matrix based on the deep space-time fusion network to generate a nursing decision-making vector; the scheme execution monitoring module is used for collecting nursing operation data flow in real time, calculating and executing deviation degree parameters and feeding back the parameters; the abnormity intervention module analyzes and executes deviation degree parameters to trigger the dynamic decision module to execute nursing scheme reconstruction; the multi-terminal cooperation module is used for converting the nursing decision vector into an equipment instruction set and feeding back an equipment state; and the secure storage module is used for encrypting and storing the feature topological graph and the decision knowledge graph and providing model parameter version management.

The invention relates to the crossing field of a medical technology and a data processing technology, in particular to a multi-dimensional data fusion ophthalmologic operation safety check data processing system, which comprises a fusion module used for responding to a process release instruction and collecting multiple paths of equipment data streams; calculating a credibility weight based on the multi-path equipment data flow, and performing weighted fusion and time sequence calibration to obtain an intra-operative fusion data set; the calibration module is used for extracting a plurality of preset key risk parameters based on the intraoperative fusion data set to form a multi-dimensional risk feature vector, and performing feature space division and statistical analysis to generate a self-adaptive adjustment coefficient; and using the adaptive adjustment coefficient to calibrate a preset risk determination reference, calculating a comprehensive risk value and outputting a grading risk warning. According to the invention, through fusion of multi-source data, dynamic risk calibration, intelligent response and cross-stage evaluation, a closed-loop data processing flow is constructed, a structured operation safety digital file is generated, and data driving of ophthalmologic operation safety check is realized.

The invention relates to the technical field of intelligent model training, in particular to an edge computing Agent model updating method supporting incremental learning, which comprises the following steps: S1, deploying a data sensing module at an edge node to capture a new input data stream in real time; s2, calculating priority scores, and screening out high-priority data blocks; s3, activating a parameter unfreezing algorithm according to the priority score, and generating a sparse update mask only covering specified parameters; s4, extracting an unfreezing parameter subset from the Agent model according to the sparse update mask; s5, executing forward reasoning and gradient local accumulation to generate an incremental gradient tensor; and S6, fusing the incremental gradient tensor into the Agent model through a weighted moving average algorithm, and outputting an updated edge Agent model. According to the method, by introducing data optimization, parameter sparse unfreezing and sliding fusion updating mechanisms, efficient, stable and self-adaptive incremental updating of the Agent model under the condition that edge resources are limited is achieved.

The application relates to the technical field of medical care, and provides a nursing system based on electronic medical records, which comprises a heterogeneous data integration module, which obtains multi-source heterogeneous electronic medical record data from a hospital HIS system, a wearable monitoring device and a nursing operation terminal and outputs the data after cleaning and standardization; a feature correlation module, which receives the data after cleaning and standardization to generate a space-time feature correlation matrix and stores a feature topology graph; a dynamic decision module, which processes the feature correlation matrix based on a deep space-time fusion network to generate a nursing decision vector; a scheme execution monitoring module, which collects nursing operation data streams in real time, calculates an execution deviation parameter and feeds back the execution deviation parameter; an abnormal intervention module, which analyzes the execution deviation parameter to trigger the dynamic decision module to execute nursing scheme reconstruction; a multi-terminal cooperation module, which converts the nursing decision vector into a device instruction set and feeds back a device state; and a safe storage module, which encrypts and stores the feature topology graph and a decision knowledge graph and provides model parameter version management.

A computing system may receive a data stream from a plurality of data sources of an edge environment. The computing system may identify a confidence score associated with each data stream received from the plurality of data sources. The computing system may, based on the associated confidence score, determine a confidence level of each data stream received from the plurality of data sources. The computing system may select the data streams received from the plurality of data sources having a first specified confidence level. The computing system may perform a data fusion process on the data streams having the first specified confidence level to generate a fused data stream that is an aggregation of the data streams having the first specified confidence level.

This invention discloses a circuit board surface mount correction method, apparatus, system, and medium. The circuit board surface mount correction method includes: acquiring basic accuracy data of the bare circuit board state and solder paste association data after solder paste has been printed on the circuit board; generating a D data stream based on the basic accuracy data, the solder paste association data, and corresponding component information, and sending the D data stream to a pick-and-place machine and a pre-reflow AOI (Automated Optical Inspection) unit; the pick-and-place machine completing component placement based on the D data stream; the pre-reflow AOI acquiring placement result information for each component on the circuit board after placement, and calculating offset data for each component based on the D data stream to generate an H data stream; and the pick-and-place machine performing surface mount correction based on the H data stream. This circuit board surface mount correction method achieves the effect of improving component placement correction accuracy and yield.

The application discloses a kind of bridge construction monitoring data anomaly identification method, system and related equipment, the first data stream of this scheme is filtered, then the statistical characteristic value of several is calculated and extracted to the data stream after filtering, and form a vector based on several statistical characteristic values, form one-dimensional feature vector X;Then the statistical characteristic value of determined feature vector X is selected according to the new feature vector Y of several corresponding characteristic values according to pre-set standard;Then the SVDD model single-class support vector machine is used, the basic classification library is called, and the feature vector Y after dimension reduction is identified, to determine whether it belongs to abnormal data;Finally, the binary classification support vector machine based on SMO algorithm is used, the basic classification library is called, and the data determined as abnormal is judged again.The scheme given by the application can realize high-precision construction monitoring abnormal data identification, improve the early warning efficiency, and can effectively avoid false alarm and false alarm.

The invention discloses a short-term imminent forecast evaluation system and a parallel calculation optimization method, and belongs to the technical field of short-term imminent forecast evaluation. Comprising a data interface module used for receiving a forecast data stream from a forecast model and an observation data stream from an observation network in real time; a dynamic evaluation module which is in communication connection with the data interface module and is used for calculating a group of multi-dimensional evaluation indexes reflecting prediction error space-time distribution characteristics based on the forecast data stream and the observation data stream; the resource decision module is in communication connection with the dynamic evaluation module and is used for receiving the multi-dimensional evaluation indexes and generating a resource scheduling instruction for the next calculation period based on a preset strategy mapping rule; parallel computing clusters; and an optimization scheduling module. According to the short-term and imminent forecast evaluation system and the parallelization calculation optimization method, the problem of resource distribution rigidity caused by disjunction of evaluation and calculation in a traditional system is solved, an intelligent feedback closed loop from evaluation to optimization is constructed, and the overall use effect is good.

A method for merging buffers and associated operations includes receiving a compute graph for a reconfigurable dataflow computing system and conducting a buffer allocation and merging process responsive to determining that a first operation specified by a first operation node is a memory indexing operation and that the first operation node is a producer for exactly one consuming node that specifies a second operation. The buffer allocation and merging process may include replacing the first operation node and the consuming node with a merged buffer node within the graph responsive to determining that the first operation and the second operation can be merged into a merged indexing operation and that the resource cost of the merged node is less than the sum of the resource costs of separate buffer nodes. A corresponding system and computer readable medium are also disclosed herein.

The invention relates to the technical field of information retrieval, in particular to an intelligent data processing method and system for a data visualization platform, and the method comprises the following steps: calculating the data activeness of each data segment based on an input data stream of the data visualization platform, screening the data segments with the data activeness higher than a preset threshold value, and generating a preliminary interest data subset. According to the method, activeness calculation and screening are carried out on each data segment in the input data stream, high-value data segments are actively identified to serve as a preliminary interest data subset, a unique identifier is generated in combination with starting and ending position information in the original data stream, and a description data unit is established, so that the original data has the capability of being accurately positioned and traced; the addressability and the tracing accuracy of the data in the retrieval process are improved; the unique identification and the position information are structurally integrated, a retrieval-oriented positioning structure is constructed, and the problem of access redundancy caused by unclear structure in traditional information storage is solved.

The invention provides a millimeter-wave radar boiler heating surface tube crack detection method based on GPU parallelization, and the method comprises the steps: constructing a three-layer parallelization frame integrating data flow, calculation and control, and employing the large-scale parallel calculation capability of a GPU to detect the crack of a heating surface tube of a millimeter-wave radar boiler. The whole-process acceleration processing from preprocessing, range profile reconstruction, feature extraction to intelligent identification is carried out on massive echo signals acquired by the millimeter wave radar, so that high-precision and high-real-time online detection and positioning of the heating surface pipe wall fine cracks are realized in a complex environment in a boiler with strong interference; finally, the detection efficiency is improved, meanwhile, the technical effect of high detection confidence is kept, and powerful technical support is provided for safe operation and preventive maintenance of the power plant boiler.

A method for data transmission in a system includes a) providing at least one piece of data information, b) modulating a carrier signal using the at least one piece of data information to form a modulated signal, c) transmitting the modulated signal, d) receiving the light signal through a camera, which captures an image, sequentially in time, as image rows or image columns, e) determining a first piece of image information and at least one second piece of image information from the image through row-wise or column-wise readings, f) forming a data stream by concatenating the first and the at least one second piece of image information, and g) calculating the signal frequency from the data stream and identifying the at least one piece of data information from the signal frequency.

Methods and systems for optimizing signaling with traffic detection function are described herein. According to an implementation, a computing device, e.g., a session management function (SMF) in a wireless communication network, may receive, from a user plane function (UPF), a first message reporting detection of a first data flow from a user equipment. The computing device may configure a bearer to carry the first data flow based on a traffic flow template (TFT) and a quality of service (QOS) assigned by a policy control function (PCF). In some examples, the computing device may receive, from the UPF, a second message reporting detection of a second data flow from the user equipment. The computing device may update the TFT and use the QoS previously assigned to the bearer to carry the second data flow.

A method for reducing latency and increasing throughput in a reconfigurable computing system includes receiving a user program for execution on a reconfigurable dataflow computing system, comprising a grid of interconnected compute units and grid of memory units. The user program includes multiple tensor-based algebraic expressions that are converted to an intermediate representation comprising multiple stages. Each stage includes one or more logical operations executable via dataflow through compute units, and each stage is preceded by and followed by a stage buffer, each stage buffer corresponding to one or more memory units. The method includes detecting a final joining stage that consumes a first and second dataflow path, having a first and second total stage buffer depth, respectively, and balancing the first and second total stage buffer depths via tuning or inserting a whole stage buffer, wherein a total associated PMU cost is minimized.

The invention discloses a fluid pulsation accelerator architecture for reinforcement learning of a large language model, and belongs to the technical field of artificial intelligence hardware acceleration. The system aims at solving the problems that when a current universal processor executes an RLHF working load, an instruction driving normal form is not matched with data flow calculation, fixed parallel granularity cannot adapt to a dynamic load, and the efficiency is low due to the fact that a framework does not sense data statistical characteristics. The core of the system is a liquid systolic array calculation fabric capable of being dynamically reconstructed, execution is triggered through data flow, and kernel boundaries and instruction overhead are eliminated. According to the system, a global lookup table subsystem is fused, and nonlinear calculation is optimized through merging query and parallel lookup by using data distribution prior; and a dynamic scheduling unit is configured, and elastic parallelism is realized by adopting a resource allocation algorithm supporting work stealing. According to the method, the throughput and the energy efficiency of RLHF training and a large language model reasoning stage can be remarkably improved, and a new design normal form is provided for a next-generation AI special computing architecture.

The invention discloses a service node quality detection method and apparatus, and a computing device. The method comprises the following steps: identifying candidate data streams with abnormal playing, and calculating an original stream frame loss rate and a transcoding stream frame loss rate of the candidate data streams; if the original stream frame loss rate is smaller than a first threshold value and the transcoding stream frame loss rate is larger than a second threshold value, determining the candidate data stream as a node abnormal data stream, and calculating a transcoding stream output-input fluctuation ratio of the node abnormal data stream; if the output-input fluctuation ratio of the transcoding stream is smaller than a third threshold value, determining a stream pushing node of the node abnormal data stream as a quality abnormal node; and if the output-input fluctuation ratio of the transcoding stream is greater than or equal to a third threshold value, determining that the transcoding node of the node abnormal data stream is a quality abnormal node. According to the scheme, the quality detection of the service node can be realized without depending on the internal monitoring data of the service node provided by a third party, the application range of the scheme is expanded, and the quality abnormal node and the node abnormal data flow can be accurately and quickly positioned.

A device may include: a bed. A sensor may be configured to configured to: sense pressure applied to the bed; and transmit a pressure stream. A plurality of temperature sensors arranged across the bed can be configured to: sense temperature in a particular location of the bed; and transmit a datastream of the sensed temperature. A computing system comprising can be configured to: receive the pressure stream; receive each of the datastreams; identify an apnea event; generate a sequence of heart rate values for the user for cardiac activity; generate a sequence of skin temperature values for the user; determine a change in heart rate; determine a change in skin temperature within the second time window; determine a relationship between the change in heart rate and the change in skin temperature; and determine a severity score for the apnea event based on the relationship.

A method for data transmission in a system includes a) providing at least one piece of data information, b) modulating a carrier signal using the at least one piece of data information to form a modulated signal, c) transmitting the modulated signal, d) receiving the light signal through a camera, which captures an image, sequentially in time, as image rows or image columns, e) determining a first piece of image information and at least one second piece of image information from the image through row-wise or column-wise readings, f) forming a data stream by concatenating the first and the at least one second piece of image information, and g) calculating the signal frequency from the data stream and identifying the at least one piece of data information from the signal frequency.

The invention relates to the technical field of deep foundation pit engineering safety monitoring, in particular to a deep foundation pit supporting structure stability evaluation and early warning system which comprises a multi-source data and construction event fusion module used for forming a standardized data stream according to acquired multi-source sensor data and construction events; the structure adaptability load quantification module is used for calculating a structure adaptability load index based on the standardized data stream; the steady-state bearing margin attenuation rate evaluation module is used for solving a steady-state bearing margin attenuation rate according to the structural adaptability load index; the trend prediction and grading early warning module is used for generating a grading early warning signal based on the steady-state bearing margin attenuation rate; according to the method, original one-sided and isolated monitoring indexes are converted into accurate quantification of structure accumulation and comprehensive loads, and the defect that evaluation is one-sided in a traditional method is overcome.

The present invention relates to an adaptive and reconfigurable dataflow computing system and method. The dataflow computing includes a dataflow-driven core to dynamically reprogram itself based on computational needs during runtime. In this system, an application program is broken down into a dataflow graph with functional units by a compiler, and an instruction dispatcher allocates and runs available processing elements to execute all functional units in the dataflow graph to perform distributed or parallel processing. The dynamic allocation of processing units can be adapted during runtime.

The invention provides a proxy encrypted traffic identification method based on context multi-stream association, which comprises the following steps: collecting a mixed traffic sample, and extracting a data stream; performing double-label calibration and statistical feature extraction on each data stream; generating a fingerprint vector for each data stream through a random forest algorithm, and calculating a target website matching index; depicting a data flow time relation based on context semantic recognition; calculating a website indication coefficient of the data flow based on the data flow extracted by multivariate feature analysis; generating a serialized fingerprint; calculating the importance score of the serialized fingerprint through the website indication coefficient of the data stream; and completing iterative training of a random forest classification model according to the obtained fingerprint vector of the data stream, the target website matching index, the serialized fingerprint of the data stream and the importance score, and completing traffic identification through the trained model. According to the method, context multi-stream association traffic identification based on machine learning can be realized under the condition that proxy encryption traffic is not directly cracked.

The invention relates to a method for detecting anomalies in a data stream. The method including a training step comprising, based on at least one reference data stream, computing a value of at least one predetermined non-conformity feature, thereby obtaining a conformity index; for each reference data stream, computing a reference behavioral dataset including, for at least one predetermined behavioral feature, a value of said behavioral feature computed based on said reference data stream, The training step also includes, for each reference data stream, generating a respective reference augmented behavioral dataset including the respective reference behavioral dataset and, for each non-conformity feature, a respective deviation value equal to zero; and training an artificial intelligence model based on each reference augmented behavioral dataset, each reference augmented behavioral dataset being associated with information indicative of the absence of anomaly in the corresponding reference data stream.